Author Topic: Replacing fried component on Encoder PCB...don't know what it is  (Read 777 times)

0 Members and 1 Guest are viewing this topic.

Offline Michael86Topic starter

  • Newbie
  • Posts: 5
  • Country: ca
Replacing fried component on Encoder PCB...don't know what it is
« on: November 24, 2024, 07:58:09 pm »
Hello,

Attachment descriptions:
3x Datasheets
1st image shows the PCB before components were removed (other than the missing component in question). As well as closeups of the chip identification markings
2nd image shoiws the bare PCB front/back with the exception of the two resistors still attached. The bottom I've traced the circuits. I apologize for using a crazy color scheme for PWR / GRD.
3rd image shows and image from a google search (aliexpress russia). Main point here is that it shows the original ceramic part. It appears to be the same thickness as the original part.


I'm trying to repair an Encoder for a wheel balancer and am hoping for some assistance. I like to play around with electronics but this is not my specialty. Please keep that in mind.

Basically I got a really good deal on the unit because the previous owner couldn't spare the time to try to repair the encoder board. The motors bearings were going and the heat from the bad bearings destroyed one of the SMD ceramic components near the Optical sensor (current theory anyways). The issue is that it had no markings at all. The previous owner had someone try to replace it. They believed it was a capacitor and apparently tried a number of different values but were unsuccessful with the repair – though they didn't use a scope at all and didn't have access to the shaft with the slotted ring for testing.

The PCB part is not available so replacing the entire unit is not an option.

I've removed most of the components to come up with an understanding of the circuit – yes...2 pads were torn and I'll have those repaired by a professional. I'm hoping to test the chips to understand if they are working or if they are also in need of replacing due to heat, but I don't know how to test them. The chips are listed below:

74HC14D – Philips NXP – Nex inverting Schmitt trigger
24C32 – believe it is Atmel
HEDR-8000 HP8K – Agilent Technologies (going off image of device and HP8K marking)

The two resistors test great and are both 100kOhm +/- 1%

The ceramic SMD capacitor measured (removed from circuit) at 106nF so I'm assuming it's a 100nF Capacitor.

The part of concern is the unknown part. It's a ceramic part, assuming its a capacitor but unknown value. It goes across the GND and VCC pins (Pin 4/5) of the HEDR-8000 which I believe is the 2K4 model.

Notes:
1) The 4 pin connector is not in use when installed in the machine – My assumption is that it's simply a port to diagnose encoder issues. Vcc and Ground are directly connected to it (assume to test for power to the unit). The two other pins are connected to the 10 pin ribbon cable but not connected to the circuit at all. So I believe one would be the A output and the other the B output. But those are just thoughts since they don't directly connect to Pins A/B of the optical sensor. I plan on hooking my scope up to the 4 pin connector after trying to replace the unknown component to observe the output. If it doesn't show, then I'll directly connect to Pin A / Pin B of the optical sensor with clips.

2) I have access to an identical working machine with identical parts with the exception of a different encoder board (newer) and different ring on the shaft (shinny ring – plastic I believe). The new encoder has a lens on it as well.

My board: eap0201d50a
New board: EAA0344G00A

I've also found reference to 6726387 being a replacement for mine. That appears to be the same as the new board but without the lens. My going theory is that the input (5v) as well as the outputs from the PCB are the same since the main PCB's that the 10 pin cable goes to are identical. The unknown is whether or not the plastic band / metal band have the same number of Bar/Window pairs and whether it would work. Ie: the new board might work if I remove the LENs. That being said, I had the chance to measure the capacitors on a "newer board" (in circuit only though) and all three of it's "capacitors" measured roughly:

Resistance:
350kOhm in one direction and 625kOhm in the other.

Capacitance:
350mF in one direction and 625mF in the other.

I know this is not a good test since it's in circuit. I didn't have the option of removing them.


Questions:

1) Anyone have enough knowledge of these types of encoders or from the datasheet of the chips/optical sensor to suggest the proper replacement / type for the missing component? Or suggestions on what to do? I assume it's a capacitor but not sure.
2) What kind of tests can I do on these specific chips to determine if they are good or not? Or should I just do a full test at the end and if it works...it works.
3) Expecting 90 deg phase shift between A/B on scope (with A leading). The input on this PCB is 5v but what are the expected outputs on this circuit? Also 0-5v on the scope from Channel A / B?
 

Offline DavidAlfa

  • Super Contributor
  • ***
  • Posts: 6402
  • Country: es
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #1 on: November 24, 2024, 08:06:32 pm »
Being connected between vcc and gnd, can only be a decoupling cap.
Try a 100nF - 1uF cap.
Hantek DSO2x1x            Drive        FAQ          DON'T BUY HANTEK! (Aka HALF-MADE)
Stm32 Soldering FW      Forum      Github      Donate
 
The following users thanked this post: Michael86

Offline HwAoRrDk

  • Super Contributor
  • ***
  • Posts: 1607
  • Country: gb
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #2 on: November 25, 2024, 12:20:51 am »
The 74HC14 is simply being used to invert and buffer the A and B signals coming from the HEDR-8000. Each signal gets inverted once with a single inverter, then the output from that goes in parallel to another pair of inverters (which will inverter the signal back again to its original polarity), whose output are again paralleled and go out to the pin header. I suspect they're paralleling a pair of inverters for output so that the signal has a strong drive strength. Possibly it passes over a long cable harness, or whatever is receiving needs a low-impedance signal.

The 24C32 is an I2C 4KB EEPROM, and most probably just used for identification and/or serialisation of the module. It should have no impact on the sensing function of the module other than perhaps the main unit may refuse to operate if it can't interrogate the EEPROM to find out whether the encoder is present and/or what it is.

You can test the EEPROM by hooking up the VCC, GND, SDA, SCL pins on the header to something like a Raspberry Pi or Arduino and trying to read out its contents over I2C. From the way the A* pins are wired, the slave address for it should be 0x54.

The two resistors test great and are both 100kOhm +/- 1%

These are pull-up resistors for the I2C SDA and SCL lines. Bit odd to have such a high value of 100k, as normally they're <10k, but maybe the main unit has its own pull-ups, and these are just fail-safes.

The part of concern is the unknown part. It's a ceramic part, assuming its a capacitor but unknown value. It goes across the GND and VCC pins (Pin 4/5) of the HEDR-8000 which I believe is the 2K4 model.

I would say it's almost certainly just a decoupling capacitor, probably of same value as the other one - 100nF.

The datasheet for the HEDR-8000 doesn't mention the need for any other external components apart from a 220-ohm current-limiting resistor for the LED (which curiously is not featured on the PCB at all!), so it's unlikely to be anything else.

2) What kind of tests can I do on these specific chips to determine if they are good or not? Or should I just do a full test at the end and if it works...it works.
3) Expecting 90 deg phase shift between A/B on scope (with A leading). The input on this PCB is 5v but what are the expected outputs on this circuit? Also 0-5v on the scope from Channel A / B?

I would try to do an end-to-end test. It should be fairly simple with a 'scope and multimeter to determine whether the module is functioning. You should be able to do it on the bench by just providing a 5V power supply. I would go about it as follows:

1. Check whether the LED in the encoder is illuminating. Jump between VLED and 5V with a 220-ohm resistor, and see if it lights. Datasheet doesn't say whether it might be a non-visible infrared LED, so if no light appears, check with camera (e.g. smartphone) to see if it lights - IR LEDs usually can be seen to glow dim purple-ish colour on camera screen. If LED is bad, you need a new HEDR-8000.
2. With the 5V powered and the LED illuminated, try to replicate the encoder wheel pattern in some kind of to-hand medium (e.g. cut some holes in paper - dimensions given in the datasheet) and see what signals you get out of the A and B pins of the HEDR-8000. The datasheet says they should have a high level at least 2.4V. No signal? Again, possibly bad HEDR-8000.
3. If you get what's expected there, then move to probe the 74HC14 input pins and see if the same signal appears. Then move to the 74HC14 outputs and check again for a signal. No output signal? Replace the 74HC14. Outputs from the inverter should have a high level around 4-5V.
4. Check the EEPROM functions and seems to contain valid data - as described above.

Other than those things, there's not really much else to go wrong. This module is pretty simple.
« Last Edit: November 25, 2024, 12:24:24 am by HwAoRrDk »
 
The following users thanked this post: Michael86

Offline Michael86Topic starter

  • Newbie
  • Posts: 5
  • Country: ca
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #3 on: November 25, 2024, 01:42:08 am »
For the output pins of the 74HC14, I take it that one output would be Pins 6, 8 (Bridged by PCB trace) and the other pins would be Pins 10, 12 (Bridged by PCB trace). I assume that's the part that you mentioned the parallel aspect and inverting of the signal / strengthing. The cable harness is quite long - at least a meter.

I ended up pulling a bunch of SMD caps off scrap boards and found mostly 4-10 microFarads BUT one of the last measured at 110 nF! I put the board back together after the first response (DavidAlfa's) to try caps between 100-1000nF and then saw the second answer (HwA0RrDk's). I decided to try as is and it worked! The first time... Then it didn't work. I manually rotated the motor and kept trying and kept getting "Loose hub" on the display, but then I got it to run again for a bit. I have a feeling that there is some bearing adhesive underneath the ring reflecting / interrupting the pickup.

Next I plan on hooking up the scope next as well as doing some of the other tests that you recommended on the bench. Even if it had worked I want to under stand what's going on.

I really appreciate the quick answers! And thank you for all that detail HwAoRrDk I'll update again once I have some time to play around again on this project.
 

Offline themadhippy

  • Super Contributor
  • ***
  • Posts: 3264
  • Country: gb
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #4 on: November 25, 2024, 02:05:07 am »
Quote
The 74HC14 is simply being used to invert and buffer the A and B signals
it also a pretty common method of reducing contact bounce
 

Offline Michael86Topic starter

  • Newbie
  • Posts: 5
  • Country: ca
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #5 on: November 26, 2024, 05:30:15 pm »
Update:

I tried some of the testing that was suggested.

First test was the 220 Ohm resistor to +5v for the LED on the Optical sensor. The sensor lit up perfectly (Red).

Next I tried to see what I got out of the encoder on the bench at Pins A and B. I read voltages just below 5v, then measured at the output pins and got 5v outputs. Instead of trying to replicate the wheel I instead soldered two wires to the output pins on the pin side of the 10 pin connector. These are marked in orange / yellow on the new picture (Image 1).

Test Points (Image 1)
Red is +5v
Black is GND
Yellow is one Output Channel (not sure which probe was connected)
Orange is the 2nd Output Channel (not sure which probe was connected)

I then connected the PCB and connected my scope to the Orange/Yellow wires and the GND point from the 4 pin connector and ran a bunch of tests trying to capture the waveform.
The waveform that I was expecting to see (if it was functioning properly) is attached in image 2. Basically 0-5v square waves 90deg phase shift split into 4 equal intervals.

Now to the results.

Image 3 shows my first test which started then errored out. Not exactly what I was hoping to see. It’s going 0-5v but the period varies and is definitely holding at 5v much longer than the low periods. Also the overlap between the two channels does not appear to be 90 deg (hard to see in this image though).

Image 4 ran the best and did not seem to error out. But the low points are very quick compared to the high points which are much longer. Ie: not 4 equal quadrants. It does appear that the overlap is at 90deg (ch1 leads ch2)

Image 5 shows some interesting variations in the length of some of the results – which appear to be missed pulses. Also quite a bit of ripple showing.

Image 6 shows Ch1 and the +5v to the PCB. I wanted to ensure that the voltage remained constant to the PCB during the test. It was constant at the +5v. I didn’t test the voltage to the LED. I guess there is a possibility that the LED flickers on / off during the test.

Image 7 shows the code wheel in a few different positions. I’ve circled the exterior cutouts in one image and put an arrow pointing to some silvery adhesive. But they are on the outside of the wheel. Could those reflect the red LED / interfere with the signal? I anticipate no and that it’s just the center strip that matters. But asking because I don’t know. Also, it’s hard to see, but zooming in on the middle image (the one with the red circle) in the center more to the left side you can see what looks to be some sort of substance below the windows. Could that affect readings?

I should also note that I found 1 spot on the shaft where the system usually starts and runs for a full cycle or runs then errors out, but not 100% of the time. Most other spots it immediately throws the “loose hub” error.

So basically…What do you think is going wrong? The results are not consistent with what I’m seeing on the scope. Could the optical sensor be failing? Should I connect a wire directly to the A / B outputs on the sensor and read those rather than the outputs of the 74HC14?

Or is this solely something to do with the ring on the shaft? Would the substance under the ring cause intermittent issues like this?
Also, any thoughts on why the +5v period is so much longer than the drops? If it wasn’t doing that I would lean towards the board being great and the shaft wheel being the issue but the high points are so long.

I haven’t done anything with the 24C32 Chip since currently the wave output has errors so I believe that is the issue.

Thanks

 

Offline Michael86Topic starter

  • Newbie
  • Posts: 5
  • Country: ca
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #6 on: November 26, 2024, 05:31:48 pm »
Didn't load image 1 for some reason
 

Offline Michael86Topic starter

  • Newbie
  • Posts: 5
  • Country: ca
Re: Replacing fried component on Encoder PCB...don't know what it is
« Reply #7 on: December 30, 2024, 03:32:07 pm »
Hello,

I've been diagnosing this in my spare time and I have a few more measurements / questions now based on my findings.

First, I took the encoder ring apart to determine what the reflective substance was underneath the ring. It was some adhesive that had been used to secure the ring to the shaft – so most likely nothing of concern and I just made a ton more work for myself for nothing.

While I had the encoder strip apart I scanned it into a PDF which I’ve uploaded. This brings me back to my previous question on the output not looking like a typical encoder output. From the PDF as well as the image of the ring, the window to bar ratio doesn’t appear exactly 50% but it’s most definitely NOT 3.9:1 (5.4ms / 1.4ms) window to bar ratio as what is being seen in the output. (I’m assuming that the 0V is the window and 5v is the bar reflection)

I’m back to thinking something is wrong with the encoder circuit and can only see 4 potentials as being the source of the problem but don’t understand them enough to know my best course of action. Based on the results below I believe the Schmitt trigger is working appropriately and just cleans the signal up since the inputs are basically identical.

Some images to support my concern:
1 – Encoder Strip Scan
1b – Encoder Strip Image
(Ch1 is the Schmitt trigger output, Ch2 is the raw A or B input to the Schmitt trigger)
2 - Encoder Channel A
3 – Encoder Channel B Overview
   Ch1 period: 6.760ms
   Ch1 Pk-Pk: 5.2V
   Ch1 Width: 5.361ms
   Ch2 Pk-Pk: 5.04V
   Ch2 Width: 5.361ms
4 – Encoder Channel B Period
   Ch2 period: 6.760ms
5 – Encoder Channel B Measured Period
   Ch1 5v: 5.4ms
6 – Encoder Channel B Measured Period
   Ch1 0v: 1.4ms
7 – Service Manual Nullmark

Basically the four options that I can see are:

1)   The capacitor that I placed is affecting the outputs. It was previously mentioned that it is most likely a decoupling capacitor – but can this affect the output of the Chip? Ie modify the ratios of the windows to bar by messing with the internal voltage? This is a simple thing to test, my issue is that I just don’t have any on hand at the moment so will have to order if that is suspected.
2)   The chip is bad even though it’s functioning.
a.   My next issue if this is suspected is exactly what HEDR-8000 chip is it.
i.   Either HEDR-8000 or HEDR-8100
ii.   I assume it’s the HEDR-8000 as there are approximately 55 lines per inch (assuming 1 line is a window/bar pair) and the HEDR-8000 is rated for 70-75 lines/inch.
iii.   Tried to find the “Application Note 1079: Design and Performance Considerations with HEDR-8000” but can’t find it.
3)   The encoder strip is not reflective enough – not really considering this at the moment as it’s extremely consistent in its outputs being the ratio shown above rather than varying randomly.
4)   I took a closer look at the board and realize that I didn’t get the encoder to sit flush on the board so it’s on a slight angle to one side. I plan on fixing this regardless but I’m curious as to if anyone knows if this could create an issue like this. I would imagine that it would make it not work and that since it’s working this isn’t the issue but I’m curious nonetheless.

Then I guess my final question is what size should the nullmark “window” be in relation to the bars? I can’t see any recommendation on the datasheet and since I took it apart and put it back together I believe it’s slightly too large now as the machine is not picking up on the nullmark as it previously did. I plan on redoing it again but was wondering if anyone had any knowledge on that.




Below is some more information from the service manual on the function of the EEPROM and how the encoder is used for position / angular reference:

Beside the EEPROM on the Controller board, there is a second non-volatile memory soldered to the Optoelectronic unit. Both EEPROMs should hold the same data: machine model, adjustment data, counts of counters, selected modes of operation and stored error-codes. Having two non-volatile memories becomes beneficial when the Controller board or the electronic box has to be replaced.

An intended irregularity in the arrangement of the mirrors is detected with the main shaft rotating at constant speed, serving as absolute reference. On detection of the reference, the position counter is set to zero thus providing absolute angular position of the main shaft

If the electronic unit fails to detect the zero reference and neither exchanging the Optoelectronic unit nor cleaning the polygon sleeve proofed successful, the following measure may help. Turn the main shaft until the solder joint of the code sleeve becomes visible. Wrap some soft cleaning paper round your forefinger and press the code sleeve slightly, so it will set in a different way. The red LED of the reflective encoder IC does not light up immediately after power on, since voltage is applied under program control! Is there a defective polygon ring, the complete vibratory assembly has to be exchanged. The main shaft cannot be removed from the vibratory tube, as the ball bearings are glued-in.

ENCODER The encoder disk is built onto the shaft. It cannot be adjusted and can only be replaced by replacing the vibratory member. The new incremental encoder is fitted in the vibratory tube and consists of a reflective slotted sleeve which is mounted on the main shaft and the optoelectronic unit. To prevent dirt and light entering, the opening in the vibratory tube must be sealed with black adhesive tape. A red visible LED and four light detectors are fitted in the encoder part of the optoelectronic unit behind the lenses. Part of the light is reflected back from the webs of the slotted sleeve to the encoder part and focussed by the lens, such that the web-slot pattern of the sleeve is mapped on the four light detectors. Two light detectors are connected to one amplifier in the encoder part. The difference in brightness between the detector pairs determines the instantaneous output states of channels A and B. To exclude interference from extraneous signals and to guarantee reliability the two signals are amplified by an IC. One slot in the sleeve is wider than the other 255 slots. Therefore the absolute angular position of the main shaft can be determined at constant rotating speed. The surface of the slotted sleeve must be clean and shiny, the slots must have a dull black background. Should a dirt particle have settled on a web or in a slot, it can be lifted off of the slotted sleeve with self-adhesive tape by applying it onto a strip of strong paper so that half of the tape is on the paper and the other half overhanging. CAUTION! If the slotted sleeve is twisted relative to the main shaft when being cleaned, the step compensation of residual shaft unbalance must be performed with F/P 84. A defective slotted sleeve cannot be replaced in the field because the ball bearings of the main shaft are pressed in.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf